Estimation of human arm motion based on sEMG in human-robot cooperative manipulation

Abstract

Human-robot cooperative manipulation receives more and more attention in recent years. Accurate motion information of the human arm can make the robot work alongside the humans and adapt to their behavior and needs. In this paper, we focused on the estimation of human arm motion based on the surface electromyography (sEMG) signals., which were detected and acquired by electrodes placed on the skin overlying the muscle. The integration of correlation analysis and modified linear discriminant analysis (LDA) was used to reduce the dimensions of sEMG features. The proposed method was verified to be effective by comparing to compared methods through human-robot cooperative crabstick sawing experiment. The human arm motion recognition rate derived by the proposed method was approximate 90%. The crabstick sawing task can be realized by the human-robot cooperative system.